[_yvjs]

> 90% of genetic variance in human races exists in the African continent

Counting neutral variation (the vast majority of genetic variation). You can get plenty of variation in non-coding DNA from drift without any effect on the variance in a trait. In fact, you rarely actually get effects on highly polygenic traits from drift without selection (imagine random directions chosen independently for each of the relevant alleles, and think in terms of the law of large numbers).

If you're getting this from Oded Galor or Deidre McCloskey, they're generally smart people but don't understand population genetics well.